Résumé
Despite notable therapeutic advances that have improved the survival of multiple myeloma (MM) patients, development of drug resistance remains a major problem. Transcriptomic analysis provides an opportunity to dissect the complexity of tumors, including the surrounding microenvironment, which has a significant impact on MM tumor progression and patients' response to treatment, as demonstrated by the effectiveness of immunomodulatory therapies. To improve the tailoring of targeted and immune based therapeutic strategies, it is crucial to decipher the tumor-immune microenvironment profile in MM patients.
We used a multi-omics data integration approach, including RNAseq-based gene expression for MM tumor cells (MMCs) (n=196) and for the corresponding purified tumor microenvironment (TME) (n=124), single nucleotide variant data from whole exome sequencing of MMCs (n=100), deconvolution of TME immune subtypes (n=124), and relevant clinical metadata. This allowed us to thoroughly characterize both the tumor and its TME.
Applying Multiomics Factor Analysis (MOFA) identified 13 principal factors capturing the diversity of the disease, encompassing both MM malignant and microenvironment components. Factors 1 and 2 specifically represented the heterogeneity in the TME, incorporating gene expression profiles and deconvolution-based immune cell subtypes. On the other hand, Factors 6, 9, 11, and 12 were specific to variations in the gene expression profile of the malignant component.
Importantly, the analysis of the variables linked to these main factors uncovers not only genes with established functions in myeloma but also novel candidates with potential importance.
Based on the top-weight in the gene expression and deconvolution data, Factor 1 showed a positive association with MMP8 expression by TME cells and a negative correlation with the presence of CD8 positive T cells in the TME. Factor 2, on the other hand, exhibited positive alignments with the expression of BHLHA15, TNFRSF17 (BCMA), KLF15, and IGF1 by TME cells.
Regarding Factor 6 and Factor 11, which capture tumor fraction heterogeneity, Factor 6 showed alignment with the expression of cancer testis antigens ( MAGEA6, MAGEA3), CXCL12, and KLF4. Meanwhile, Factor 11 was associated with the expression of genes such as SULF2, as well as CRP levels, percentage of CD138+ cells, and the percentage of cells in cell cycle phase S.
Next, we investigated the predictive potential of the latent factors inferred by MOFA in models of clinical outcomes. Interestingly, 6 out of the 13 factors identified by MOFA showed significant associations with patients' overall survival (cox regression, p <= 0.05). These factors include Factor 1 and 2, which are related to the composition of the TME, and Factor 6 and 9, which are associated with myeloma progression status, osteolysis, t(11;14), t(4;14), and MM molecular classification.
Finally, performing unsupervised clustering using the latent MOFA factors, we identified two main groups with distinct TME subtypes. These groups can be further divided into nine subgroups, each exhibiting unique TME characteristics. Notably, the TME subtypes showed correlations with essential gene expression signatures related to immune response regulation, T cell activation, and natural killer mediated immunity. Additionally, the significant presence of deconvolution-based immune cell types, including T cell CD8+, T cell regulatory, T cell CD4+ (non-regulatory), NK cells and Monocytes was observed. Moreover, the TME subtypes were associated with patients' overall survival and response to daratumumab. Patients presenting TME enriched in immune cells were associated with a significant better outcome ( p = 0.0026).
These findings highlight the importance of TME analysis in predicting treatment response and MM patients' outcome in the context of immune-based therapies. Our integrative multi-omics analysis revealed comprehensive MM heterogeneity and distinct immune subtypes that could be of therapeutic interest for personalized therapeutic approaches.
Vincent:Pfizer: Other: Financing meeting participation; BMS, Takeda: Membership on an entity's Board of Directors or advisory committees, Other: Financing meeting participation; Janssen: Honoraria, Membership on an entity's Board of Directors or advisory committees, Other: Financing meeting participation. Herbaux:AbbVie, F. Hoffmann-La Roche Ltd, AstraZeneca, Janssen: Honoraria; AbbVie, F. Hoffmann-La Roche Ltd, AstraZeneca, Janssen: Consultancy; Physician and professor of Hematology at academic center (CHU Montpellier France): Current Employment; AbbVie, Takeda: Research Funding.